IMPORT LIBRARIES AND DATASETS¶

In [1]:
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
from sklearn.preprocessing import StandardScaler, normalize
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
import os
import warnings
warnings.filterwarnings('ignore')
In [2]:
creditcard_df=pd.read_csv('D:\\data science\\Batch67 Day28\\Projects\\Project -5 Marketting Department\\Marketing_data.csv')
creditcard_df
Out[2]:
CUST_ID BALANCE BALANCE_FREQUENCY PURCHASES ONEOFF_PURCHASES INSTALLMENTS_PURCHASES CASH_ADVANCE PURCHASES_FREQUENCY ONEOFF_PURCHASES_FREQUENCY PURCHASES_INSTALLMENTS_FREQUENCY CASH_ADVANCE_FREQUENCY CASH_ADVANCE_TRX PURCHASES_TRX CREDIT_LIMIT PAYMENTS MINIMUM_PAYMENTS PRC_FULL_PAYMENT TENURE
0 C10001 40.900749 0.818182 95.40 0.00 95.40 0.000000 0.166667 0.000000 0.083333 0.000000 0 2 1000.0 201.802084 139.509787 0.000000 12
1 C10002 3202.467416 0.909091 0.00 0.00 0.00 6442.945483 0.000000 0.000000 0.000000 0.250000 4 0 7000.0 4103.032597 1072.340217 0.222222 12
2 C10003 2495.148862 1.000000 773.17 773.17 0.00 0.000000 1.000000 1.000000 0.000000 0.000000 0 12 7500.0 622.066742 627.284787 0.000000 12
3 C10004 1666.670542 0.636364 1499.00 1499.00 0.00 205.788017 0.083333 0.083333 0.000000 0.083333 1 1 7500.0 0.000000 NaN 0.000000 12
4 C10005 817.714335 1.000000 16.00 16.00 0.00 0.000000 0.083333 0.083333 0.000000 0.000000 0 1 1200.0 678.334763 244.791237 0.000000 12
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
8945 C19186 28.493517 1.000000 291.12 0.00 291.12 0.000000 1.000000 0.000000 0.833333 0.000000 0 6 1000.0 325.594462 48.886365 0.500000 6
8946 C19187 19.183215 1.000000 300.00 0.00 300.00 0.000000 1.000000 0.000000 0.833333 0.000000 0 6 1000.0 275.861322 NaN 0.000000 6
8947 C19188 23.398673 0.833333 144.40 0.00 144.40 0.000000 0.833333 0.000000 0.666667 0.000000 0 5 1000.0 81.270775 82.418369 0.250000 6
8948 C19189 13.457564 0.833333 0.00 0.00 0.00 36.558778 0.000000 0.000000 0.000000 0.166667 2 0 500.0 52.549959 55.755628 0.250000 6
8949 C19190 372.708075 0.666667 1093.25 1093.25 0.00 127.040008 0.666667 0.666667 0.000000 0.333333 2 23 1200.0 63.165404 88.288956 0.000000 6

8950 rows × 18 columns

Check details¶

In [3]:
creditcard_df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 8950 entries, 0 to 8949
Data columns (total 18 columns):
 #   Column                            Non-Null Count  Dtype  
---  ------                            --------------  -----  
 0   CUST_ID                           8950 non-null   object 
 1   BALANCE                           8950 non-null   float64
 2   BALANCE_FREQUENCY                 8950 non-null   float64
 3   PURCHASES                         8950 non-null   float64
 4   ONEOFF_PURCHASES                  8950 non-null   float64
 5   INSTALLMENTS_PURCHASES            8950 non-null   float64
 6   CASH_ADVANCE                      8950 non-null   float64
 7   PURCHASES_FREQUENCY               8950 non-null   float64
 8   ONEOFF_PURCHASES_FREQUENCY        8950 non-null   float64
 9   PURCHASES_INSTALLMENTS_FREQUENCY  8950 non-null   float64
 10  CASH_ADVANCE_FREQUENCY            8950 non-null   float64
 11  CASH_ADVANCE_TRX                  8950 non-null   int64  
 12  PURCHASES_TRX                     8950 non-null   int64  
 13  CREDIT_LIMIT                      8949 non-null   float64
 14  PAYMENTS                          8950 non-null   float64
 15  MINIMUM_PAYMENTS                  8637 non-null   float64
 16  PRC_FULL_PAYMENT                  8950 non-null   float64
 17  TENURE                            8950 non-null   int64  
dtypes: float64(14), int64(3), object(1)
memory usage: 1.2+ MB
In [4]:
creditcard_df.describe()
Out[4]:
BALANCE BALANCE_FREQUENCY PURCHASES ONEOFF_PURCHASES INSTALLMENTS_PURCHASES CASH_ADVANCE PURCHASES_FREQUENCY ONEOFF_PURCHASES_FREQUENCY PURCHASES_INSTALLMENTS_FREQUENCY CASH_ADVANCE_FREQUENCY CASH_ADVANCE_TRX PURCHASES_TRX CREDIT_LIMIT PAYMENTS MINIMUM_PAYMENTS PRC_FULL_PAYMENT TENURE
count 8950.000000 8950.000000 8950.000000 8950.000000 8950.000000 8950.000000 8950.000000 8950.000000 8950.000000 8950.000000 8950.000000 8950.000000 8949.000000 8950.000000 8637.000000 8950.000000 8950.000000
mean 1564.474828 0.877271 1003.204834 592.437371 411.067645 978.871112 0.490351 0.202458 0.364437 0.135144 3.248827 14.709832 4494.449450 1733.143852 864.206542 0.153715 11.517318
std 2081.531879 0.236904 2136.634782 1659.887917 904.338115 2097.163877 0.401371 0.298336 0.397448 0.200121 6.824647 24.857649 3638.815725 2895.063757 2372.446607 0.292499 1.338331
min 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 50.000000 0.000000 0.019163 0.000000 6.000000
25% 128.281915 0.888889 39.635000 0.000000 0.000000 0.000000 0.083333 0.000000 0.000000 0.000000 0.000000 1.000000 1600.000000 383.276166 169.123707 0.000000 12.000000
50% 873.385231 1.000000 361.280000 38.000000 89.000000 0.000000 0.500000 0.083333 0.166667 0.000000 0.000000 7.000000 3000.000000 856.901546 312.343947 0.000000 12.000000
75% 2054.140036 1.000000 1110.130000 577.405000 468.637500 1113.821139 0.916667 0.300000 0.750000 0.222222 4.000000 17.000000 6500.000000 1901.134317 825.485459 0.142857 12.000000
max 19043.138560 1.000000 49039.570000 40761.250000 22500.000000 47137.211760 1.000000 1.000000 1.000000 1.500000 123.000000 358.000000 30000.000000 50721.483360 76406.207520 1.000000 12.000000
In [5]:
#check who made one off purchase of $40761 ie maximum ONEOFF_PURCHASES
creditcard_df[creditcard_df['ONEOFF_PURCHASES']==40761.25]
Out[5]:
CUST_ID BALANCE BALANCE_FREQUENCY PURCHASES ONEOFF_PURCHASES INSTALLMENTS_PURCHASES CASH_ADVANCE PURCHASES_FREQUENCY ONEOFF_PURCHASES_FREQUENCY PURCHASES_INSTALLMENTS_FREQUENCY CASH_ADVANCE_FREQUENCY CASH_ADVANCE_TRX PURCHASES_TRX CREDIT_LIMIT PAYMENTS MINIMUM_PAYMENTS PRC_FULL_PAYMENT TENURE
550 C10574 11547.52001 1.0 49039.57 40761.25 8278.32 558.166886 1.0 1.0 0.916667 0.083333 1 101 22500.0 46930.59824 2974.069421 0.25 12
In [6]:
# Check who made cash advance of $47137
# This customer made 123 cash advance transactions
# Never paid credit card in full
creditcard_df[creditcard_df['CASH_ADVANCE']==47137.211760]
Out[6]:
CUST_ID BALANCE BALANCE_FREQUENCY PURCHASES ONEOFF_PURCHASES INSTALLMENTS_PURCHASES CASH_ADVANCE PURCHASES_FREQUENCY ONEOFF_PURCHASES_FREQUENCY PURCHASES_INSTALLMENTS_FREQUENCY CASH_ADVANCE_FREQUENCY CASH_ADVANCE_TRX PURCHASES_TRX CREDIT_LIMIT PAYMENTS MINIMUM_PAYMENTS PRC_FULL_PAYMENT TENURE
2159 C12226 10905.05381 1.0 431.93 133.5 298.43 47137.21176 0.583333 0.25 0.5 1.0 123 21 19600.0 39048.59762 5394.173671 0.0 12

VISUALIZE AND EXPLORE DATASET¶

In [7]:
creditcard_df.isna().sum()
Out[7]:
CUST_ID                               0
BALANCE                               0
BALANCE_FREQUENCY                     0
PURCHASES                             0
ONEOFF_PURCHASES                      0
INSTALLMENTS_PURCHASES                0
CASH_ADVANCE                          0
PURCHASES_FREQUENCY                   0
ONEOFF_PURCHASES_FREQUENCY            0
PURCHASES_INSTALLMENTS_FREQUENCY      0
CASH_ADVANCE_FREQUENCY                0
CASH_ADVANCE_TRX                      0
PURCHASES_TRX                         0
CREDIT_LIMIT                          1
PAYMENTS                              0
MINIMUM_PAYMENTS                    313
PRC_FULL_PAYMENT                      0
TENURE                                0
dtype: int64
In [8]:
import seaborn as sns
# Heat map for missing data  
sns.heatmap(creditcard_df.isnull(), yticklabels = False, cbar = False, cmap="Blues")
Out[8]:
<AxesSubplot:>
In [9]:
#Fill up the missing elements with mean of the 'MINIMUM_PAYMENT' 
creditcard_df.loc[(creditcard_df['MINIMUM_PAYMENTS'].isnull() == True), 'MINIMUM_PAYMENTS'] = creditcard_df['MINIMUM_PAYMENTS'].mean()
In [10]:
creditcard_df.isna().sum()
Out[10]:
CUST_ID                             0
BALANCE                             0
BALANCE_FREQUENCY                   0
PURCHASES                           0
ONEOFF_PURCHASES                    0
INSTALLMENTS_PURCHASES              0
CASH_ADVANCE                        0
PURCHASES_FREQUENCY                 0
ONEOFF_PURCHASES_FREQUENCY          0
PURCHASES_INSTALLMENTS_FREQUENCY    0
CASH_ADVANCE_FREQUENCY              0
CASH_ADVANCE_TRX                    0
PURCHASES_TRX                       0
CREDIT_LIMIT                        1
PAYMENTS                            0
MINIMUM_PAYMENTS                    0
PRC_FULL_PAYMENT                    0
TENURE                              0
dtype: int64
In [11]:
# Fill up the missing elements with mean of the 'CREDIT_LIMIT' 
creditcard_df.loc[(creditcard_df['CREDIT_LIMIT'].isnull() == True), 'CREDIT_LIMIT'] = creditcard_df['CREDIT_LIMIT'].mean()
In [12]:
()
Out[12]:
()
In [13]:
creditcard_df.duplicated().sum()
Out[13]:
0
In [14]:
sns.heatmap(creditcard_df.isnull(),yticklabels=False,cbar=False,cmap='Blues')
Out[14]:
<AxesSubplot:>
In [15]:
creditcard_df.shape
Out[15]:
(8950, 18)
In [16]:
creditcard_df.drop('CUST_ID',axis=1,inplace=True)
display(creditcard_df)
BALANCE BALANCE_FREQUENCY PURCHASES ONEOFF_PURCHASES INSTALLMENTS_PURCHASES CASH_ADVANCE PURCHASES_FREQUENCY ONEOFF_PURCHASES_FREQUENCY PURCHASES_INSTALLMENTS_FREQUENCY CASH_ADVANCE_FREQUENCY CASH_ADVANCE_TRX PURCHASES_TRX CREDIT_LIMIT PAYMENTS MINIMUM_PAYMENTS PRC_FULL_PAYMENT TENURE
0 40.900749 0.818182 95.40 0.00 95.40 0.000000 0.166667 0.000000 0.083333 0.000000 0 2 1000.0 201.802084 139.509787 0.000000 12
1 3202.467416 0.909091 0.00 0.00 0.00 6442.945483 0.000000 0.000000 0.000000 0.250000 4 0 7000.0 4103.032597 1072.340217 0.222222 12
2 2495.148862 1.000000 773.17 773.17 0.00 0.000000 1.000000 1.000000 0.000000 0.000000 0 12 7500.0 622.066742 627.284787 0.000000 12
3 1666.670542 0.636364 1499.00 1499.00 0.00 205.788017 0.083333 0.083333 0.000000 0.083333 1 1 7500.0 0.000000 864.206542 0.000000 12
4 817.714335 1.000000 16.00 16.00 0.00 0.000000 0.083333 0.083333 0.000000 0.000000 0 1 1200.0 678.334763 244.791237 0.000000 12
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
8945 28.493517 1.000000 291.12 0.00 291.12 0.000000 1.000000 0.000000 0.833333 0.000000 0 6 1000.0 325.594462 48.886365 0.500000 6
8946 19.183215 1.000000 300.00 0.00 300.00 0.000000 1.000000 0.000000 0.833333 0.000000 0 6 1000.0 275.861322 864.206542 0.000000 6
8947 23.398673 0.833333 144.40 0.00 144.40 0.000000 0.833333 0.000000 0.666667 0.000000 0 5 1000.0 81.270775 82.418369 0.250000 6
8948 13.457564 0.833333 0.00 0.00 0.00 36.558778 0.000000 0.000000 0.000000 0.166667 2 0 500.0 52.549959 55.755628 0.250000 6
8949 372.708075 0.666667 1093.25 1093.25 0.00 127.040008 0.666667 0.666667 0.000000 0.333333 2 23 1200.0 63.165404 88.288956 0.000000 6

8950 rows × 17 columns

In [17]:
len(creditcard_df.columns)
Out[17]:
17
In [18]:
creditcard_df.columns
Out[18]:
Index(['BALANCE', 'BALANCE_FREQUENCY', 'PURCHASES', 'ONEOFF_PURCHASES',
       'INSTALLMENTS_PURCHASES', 'CASH_ADVANCE', 'PURCHASES_FREQUENCY',
       'ONEOFF_PURCHASES_FREQUENCY', 'PURCHASES_INSTALLMENTS_FREQUENCY',
       'CASH_ADVANCE_FREQUENCY', 'CASH_ADVANCE_TRX', 'PURCHASES_TRX',
       'CREDIT_LIMIT', 'PAYMENTS', 'MINIMUM_PAYMENTS', 'PRC_FULL_PAYMENT',
       'TENURE'],
      dtype='object')
In [19]:
# distplot combines the matplotlib.hist function with seaborn kdeplot()
# KDE Plot represents the Kernel Density Estimate
# KDE is used for visualizing the Probability Density of a continuous variable. 
# KDE demonstrates the probability density at different values in a continuous variable. 

# Mean of balance is $1500
# 'Balance_Frequency' for most customers is updated frequently ~1
# For 'PURCHASES_FREQUENCY', there are two distinct group of customers
# For 'ONEOFF_PURCHASES_FREQUENCY' and 'PURCHASES_INSTALLMENT_FREQUENCY' most users don't do one off puchases or installment purchases frequently 
# Very small number of customers pay their balance in full 'PRC_FULL_PAYMENT'~0
# Credit limit average is around $4500
# Most customers are ~11 years tenure

plt.figure(figsize=(20,80))
for i in range(len(creditcard_df.columns)):
    plt.subplot(17, 1, i+1)
    sns.distplot(creditcard_df[creditcard_df.columns[i]], kde_kws={"color": "b", "lw": 3, "label": "KDE"}, hist_kws={"color": "g"})
    plt.title(creditcard_df.columns[i])

plt.tight_layout()
In [20]:
sns.pairplot(creditcard_df)
# Correlation between 'PURCHASES' and ONEOFF_PURCHASES & INSTALMENT_PURCHASES 
# Trend between 'PURCHASES' and 'CREDIT_LIMIT' & 'PAYMENTS'
Out[20]:
<seaborn.axisgrid.PairGrid at 0x2019424c610>
In [21]:
correlations = creditcard_df.corr()
display (correlations )
BALANCE BALANCE_FREQUENCY PURCHASES ONEOFF_PURCHASES INSTALLMENTS_PURCHASES CASH_ADVANCE PURCHASES_FREQUENCY ONEOFF_PURCHASES_FREQUENCY PURCHASES_INSTALLMENTS_FREQUENCY CASH_ADVANCE_FREQUENCY CASH_ADVANCE_TRX PURCHASES_TRX CREDIT_LIMIT PAYMENTS MINIMUM_PAYMENTS PRC_FULL_PAYMENT TENURE
BALANCE 1.000000 0.322412 0.181261 0.164350 0.126469 0.496692 -0.077944 0.073166 -0.063186 0.449218 0.385152 0.154338 0.531267 0.322802 0.394282 -0.318959 0.072692
BALANCE_FREQUENCY 0.322412 1.000000 0.133674 0.104323 0.124292 0.099388 0.229715 0.202415 0.176079 0.191873 0.141555 0.189626 0.095795 0.065008 0.114249 -0.095082 0.119776
PURCHASES 0.181261 0.133674 1.000000 0.916845 0.679896 -0.051474 0.393017 0.498430 0.315567 -0.120143 -0.067175 0.689561 0.356959 0.603264 0.093515 0.180379 0.086288
ONEOFF_PURCHASES 0.164350 0.104323 0.916845 1.000000 0.330622 -0.031326 0.264937 0.524891 0.127729 -0.082628 -0.046212 0.545523 0.319721 0.567292 0.048597 0.132763 0.064150
INSTALLMENTS_PURCHASES 0.126469 0.124292 0.679896 0.330622 1.000000 -0.064244 0.442418 0.214042 0.511351 -0.132318 -0.073999 0.628108 0.256496 0.384084 0.131687 0.182569 0.086143
CASH_ADVANCE 0.496692 0.099388 -0.051474 -0.031326 -0.064244 1.000000 -0.215507 -0.086754 -0.177070 0.628522 0.656498 -0.075850 0.303983 0.453238 0.139223 -0.152935 -0.068312
PURCHASES_FREQUENCY -0.077944 0.229715 0.393017 0.264937 0.442418 -0.215507 1.000000 0.501343 0.862934 -0.308478 -0.203478 0.568430 0.119778 0.103464 0.002976 0.305802 0.061506
ONEOFF_PURCHASES_FREQUENCY 0.073166 0.202415 0.498430 0.524891 0.214042 -0.086754 0.501343 1.000000 0.142329 -0.111716 -0.069088 0.544869 0.295030 0.243537 -0.029963 0.157531 0.082466
PURCHASES_INSTALLMENTS_FREQUENCY -0.063186 0.176079 0.315567 0.127729 0.511351 -0.177070 0.862934 0.142329 1.000000 -0.262958 -0.169207 0.529975 0.060752 0.085551 0.029590 0.250087 0.073275
CASH_ADVANCE_FREQUENCY 0.449218 0.191873 -0.120143 -0.082628 -0.132318 0.628522 -0.308478 -0.111716 -0.262958 1.000000 0.799561 -0.131168 0.132616 0.183192 0.097898 -0.249773 -0.133372
CASH_ADVANCE_TRX 0.385152 0.141555 -0.067175 -0.046212 -0.073999 0.656498 -0.203478 -0.069088 -0.169207 0.799561 1.000000 -0.066157 0.149699 0.255278 0.109185 -0.169784 -0.043421
PURCHASES_TRX 0.154338 0.189626 0.689561 0.545523 0.628108 -0.075850 0.568430 0.544869 0.529975 -0.131168 -0.066157 1.000000 0.272877 0.370832 0.095858 0.162066 0.121874
CREDIT_LIMIT 0.531267 0.095795 0.356959 0.319721 0.256496 0.303983 0.119778 0.295030 0.060752 0.132616 0.149699 0.272877 1.000000 0.421852 0.125134 0.055671 0.139034
PAYMENTS 0.322802 0.065008 0.603264 0.567292 0.384084 0.453238 0.103464 0.243537 0.085551 0.183192 0.255278 0.370832 0.421852 1.000000 0.125046 0.112138 0.106136
MINIMUM_PAYMENTS 0.394282 0.114249 0.093515 0.048597 0.131687 0.139223 0.002976 -0.029963 0.029590 0.097898 0.109185 0.095858 0.125134 0.125046 1.000000 -0.139674 0.057257
PRC_FULL_PAYMENT -0.318959 -0.095082 0.180379 0.132763 0.182569 -0.152935 0.305802 0.157531 0.250087 -0.249773 -0.169784 0.162066 0.055671 0.112138 -0.139674 1.000000 -0.016486
TENURE 0.072692 0.119776 0.086288 0.064150 0.086143 -0.068312 0.061506 0.082466 0.073275 -0.133372 -0.043421 0.121874 0.139034 0.106136 0.057257 -0.016486 1.000000
In [22]:
f, ax = plt.subplots(figsize = (20, 20))
sns.heatmap(correlations, annot = True)

# 'PURCHASES' have high correlation between one-off purchases, 'installment purchases, purchase transactions, credit limit and payments. 
# Strong Positive Correlation between 'PURCHASES_FREQUENCY' and 'PURCHASES_INSTALLMENT_FREQUENCY'
Out[22]:
<AxesSubplot:>
In [23]:
display (creditcard_df)
BALANCE BALANCE_FREQUENCY PURCHASES ONEOFF_PURCHASES INSTALLMENTS_PURCHASES CASH_ADVANCE PURCHASES_FREQUENCY ONEOFF_PURCHASES_FREQUENCY PURCHASES_INSTALLMENTS_FREQUENCY CASH_ADVANCE_FREQUENCY CASH_ADVANCE_TRX PURCHASES_TRX CREDIT_LIMIT PAYMENTS MINIMUM_PAYMENTS PRC_FULL_PAYMENT TENURE
0 40.900749 0.818182 95.40 0.00 95.40 0.000000 0.166667 0.000000 0.083333 0.000000 0 2 1000.0 201.802084 139.509787 0.000000 12
1 3202.467416 0.909091 0.00 0.00 0.00 6442.945483 0.000000 0.000000 0.000000 0.250000 4 0 7000.0 4103.032597 1072.340217 0.222222 12
2 2495.148862 1.000000 773.17 773.17 0.00 0.000000 1.000000 1.000000 0.000000 0.000000 0 12 7500.0 622.066742 627.284787 0.000000 12
3 1666.670542 0.636364 1499.00 1499.00 0.00 205.788017 0.083333 0.083333 0.000000 0.083333 1 1 7500.0 0.000000 864.206542 0.000000 12
4 817.714335 1.000000 16.00 16.00 0.00 0.000000 0.083333 0.083333 0.000000 0.000000 0 1 1200.0 678.334763 244.791237 0.000000 12
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
8945 28.493517 1.000000 291.12 0.00 291.12 0.000000 1.000000 0.000000 0.833333 0.000000 0 6 1000.0 325.594462 48.886365 0.500000 6
8946 19.183215 1.000000 300.00 0.00 300.00 0.000000 1.000000 0.000000 0.833333 0.000000 0 6 1000.0 275.861322 864.206542 0.000000 6
8947 23.398673 0.833333 144.40 0.00 144.40 0.000000 0.833333 0.000000 0.666667 0.000000 0 5 1000.0 81.270775 82.418369 0.250000 6
8948 13.457564 0.833333 0.00 0.00 0.00 36.558778 0.000000 0.000000 0.000000 0.166667 2 0 500.0 52.549959 55.755628 0.250000 6
8949 372.708075 0.666667 1093.25 1093.25 0.00 127.040008 0.666667 0.666667 0.000000 0.333333 2 23 1200.0 63.165404 88.288956 0.000000 6

8950 rows × 17 columns

OPTIMAL NUMBER OF CLUSTERS¶

In [24]:
# Let's scale the data first
scaler = StandardScaler()
creditcard_df_scaled = scaler.fit_transform(creditcard_df)
In [25]:
creditcard_df_scaled.shape
Out[25]:
(8950, 17)
In [26]:
display(pd.DataFrame(creditcard_df_scaled))
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
0 -0.731989 -0.249434 -0.424900 -0.356934 -0.349079 -0.466786 -0.806490 -0.678661 -0.707313 -0.675349 -0.476070 -0.511333 -0.960433 -0.528979 -3.109675e-01 -0.525551 0.360680
1 0.786961 0.134325 -0.469552 -0.356934 -0.454576 2.605605 -1.221758 -0.678661 -0.916995 0.573963 0.110074 -0.591796 0.688639 0.818642 8.931021e-02 0.234227 0.360680
2 0.447135 0.518084 -0.107668 0.108889 -0.454576 -0.466786 1.269843 2.673451 -0.916995 -0.675349 -0.476070 -0.109020 0.826062 -0.383805 -1.016632e-01 -0.525551 0.360680
3 0.049099 -1.016953 0.232058 0.546189 -0.454576 -0.368653 -1.014125 -0.399319 -0.916995 -0.258913 -0.329534 -0.551565 0.826062 -0.598688 -4.390474e-16 -0.525551 0.360680
4 -0.358775 0.518084 -0.462063 -0.347294 -0.454576 -0.466786 -1.014125 -0.399319 -0.916995 -0.675349 -0.476070 -0.551565 -0.905464 -0.364368 -2.657913e-01 -0.525551 0.360680
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
8945 -0.737950 0.518084 -0.333293 -0.356934 -0.132643 -0.466786 1.269843 -0.678661 1.179833 -0.675349 -0.476070 -0.350408 -0.960433 -0.486217 -3.498541e-01 1.183951 -4.122768
8946 -0.742423 0.518084 -0.329136 -0.356934 -0.122823 -0.466786 1.269843 -0.678661 1.179833 -0.675349 -0.476070 -0.350408 -0.960433 -0.503396 -4.390474e-16 -0.525551 -4.122768
8947 -0.740398 -0.185477 -0.401965 -0.356934 -0.294893 -0.466786 0.854576 -0.678661 0.760469 -0.675349 -0.476070 -0.390639 -0.960433 -0.570615 -3.354655e-01 0.329200 -4.122768
8948 -0.745174 -0.185477 -0.469552 -0.356934 -0.454576 -0.449352 -1.221758 -0.678661 -0.916995 0.157527 -0.182998 -0.591796 -1.097856 -0.580536 -3.469065e-01 0.329200 -4.122768
8949 -0.572575 -0.889033 0.042146 0.301732 -0.454576 -0.406205 0.439310 1.556082 -0.916995 0.990398 -0.182998 0.333524 -0.905464 -0.576869 -3.329464e-01 -0.525551 -4.122768

8950 rows × 17 columns

In [27]:
scores_1 = []

range_values = range(1, 20)

for i in range_values:
    kmeans = KMeans(n_clusters = i)
    kmeans.fit(creditcard_df_scaled)
    scores_1.append(kmeans.inertia_) 

plt.plot(scores_1, 'bx-')
plt.title('Finding the right number of clusters')
plt.xlabel('Clusters')
plt.ylabel('Scores') 
plt.show()

# From this we can observe that, 4th cluster seems to be forming the elbow of the curve. 
# However, the values does not reduce linearly until 8th cluster. 
# Let's choose the number of clusters to be 7.
In [28]:
kmeans = KMeans(8)
kmeans.fit(creditcard_df_scaled)
labels = kmeans.labels_
In [29]:
kmeans.cluster_centers_.shape
Out[29]:
(8, 17)
In [30]:
cluster_centers = pd.DataFrame(data = kmeans.cluster_centers_, columns = [creditcard_df.columns])
display(cluster_centers )
BALANCE BALANCE_FREQUENCY PURCHASES ONEOFF_PURCHASES INSTALLMENTS_PURCHASES CASH_ADVANCE PURCHASES_FREQUENCY ONEOFF_PURCHASES_FREQUENCY PURCHASES_INSTALLMENTS_FREQUENCY CASH_ADVANCE_FREQUENCY CASH_ADVANCE_TRX PURCHASES_TRX CREDIT_LIMIT PAYMENTS MINIMUM_PAYMENTS PRC_FULL_PAYMENT TENURE
0 -0.366639 0.333155 -0.038243 -0.244205 0.358036 -0.363598 0.991214 -0.386909 1.206355 -0.475154 -0.361158 0.187123 -0.260825 -0.217616 -0.045775 0.315794 0.257381
1 1.690200 0.396003 -0.219127 -0.157223 -0.229345 2.020153 -0.480727 -0.214389 -0.417038 1.925066 1.939912 -0.268180 1.027024 0.808261 0.528696 -0.395008 0.069894
2 -0.162873 0.392754 0.473201 0.617653 -0.016010 -0.333808 0.944809 1.880640 0.090573 -0.408068 -0.324367 0.532389 0.383794 0.100567 -0.162365 0.414080 0.262038
3 -0.701163 -2.144787 -0.311233 -0.235823 -0.302542 -0.319756 -0.557150 -0.445187 -0.440134 -0.520622 -0.376063 -0.419936 -0.175963 -0.192408 -0.256478 0.283762 0.199336
4 0.934901 0.467049 2.252953 1.707412 2.190297 -0.192704 1.160751 1.548096 1.257221 -0.310075 -0.208712 2.800594 1.235324 1.282259 0.575854 0.273318 0.333736
5 1.923051 0.337717 11.212042 10.600367 7.033118 0.419625 1.046983 1.915501 0.981334 -0.258912 0.061229 5.362438 3.044064 8.098975 1.120318 1.110132 0.310863
6 -0.336050 -0.347078 -0.289267 -0.215966 -0.286835 0.068284 -0.203078 -0.288661 -0.224549 0.308663 0.000996 -0.388117 -0.567159 -0.392680 -0.209145 0.014011 -3.203733
7 0.016269 0.403662 -0.361585 -0.246994 -0.401079 -0.090629 -0.865326 -0.409276 -0.757165 0.110146 -0.024439 -0.486185 -0.306840 -0.249619 -0.009712 -0.456880 0.272704
In [31]:
# In order to understand what these numbers mean, perform inverse transformation
cluster_centers = scaler.inverse_transform(cluster_centers)
cluster_centers = pd.DataFrame(data = cluster_centers, columns = [creditcard_df.columns])
display(cluster_centers)

# First Customers cluster (Transactors): Those are customers who pay least amount of intrerest charges and careful with their money, Cluster with lowest balance ($104) and cash advance ($303), Percentage of full payment = 23%
# Second customers cluster (revolvers) who use credit card as a loan (most lucrative sector): highest balance ($5000) and cash advance (~$5000), low purchase frequency, high cash advance frequency (0.5), high cash advance transactions (16) and low percentage of full payment (3%)
# Third customer cluster (VIP/Prime): high credit limit $16K and highest percentage of full payment, target for increase credit limit and increase spending habits
# Fourth customer cluster (low tenure): these are customers with low tenure (7 years), low balance
BALANCE BALANCE_FREQUENCY PURCHASES ONEOFF_PURCHASES INSTALLMENTS_PURCHASES CASH_ADVANCE PURCHASES_FREQUENCY ONEOFF_PURCHASES_FREQUENCY PURCHASES_INSTALLMENTS_FREQUENCY CASH_ADVANCE_FREQUENCY CASH_ADVANCE_TRX PURCHASES_TRX CREDIT_LIMIT PAYMENTS MINIMUM_PAYMENTS PRC_FULL_PAYMENT TENURE
0 801.347031 0.956192 921.498712 187.107663 734.835246 216.389570 0.888173 0.087035 0.843874 0.040061 0.784187 19.361014 3545.460071 1103.167378 757.529897 0.246079 11.861760
1 5082.483731 0.971080 535.035681 331.479330 203.673753 5215.225625 0.297412 0.138501 0.198696 0.520370 16.487298 8.043880 8231.183078 4072.979303 2096.311415 0.038182 11.610855
2 1225.469199 0.970310 2014.205154 1617.614991 396.590163 278.861158 0.869548 0.763489 0.400433 0.053486 1.035262 27.943038 5890.847444 2024.274180 485.822410 0.274826 11.867993
3 105.063837 0.369191 338.250830 201.019898 137.482244 308.328143 0.266740 0.069650 0.189517 0.030962 0.682472 4.271804 3854.224762 1176.141747 266.495312 0.236710 11.784081
4 3510.393168 0.987910 5816.674519 3426.390962 2391.725865 574.762188 0.956216 0.664285 0.864089 0.073095 1.824519 84.322115 8989.062500 5445.158665 2206.210364 0.233656 11.963942
5 5567.142164 0.957273 24957.905000 18186.875667 6771.029333 1858.844605 0.910556 0.773889 0.754444 0.083333 3.666667 148.000000 15570.000000 25178.882690 3475.059479 0.478409 11.933333
6 865.015978 0.795051 385.181720 233.977974 151.686061 1122.064941 0.408846 0.116344 0.275196 0.196911 3.255627 5.062701 2430.891398 596.373827 376.802926 0.157813 7.229904
7 1598.338053 0.972895 230.672016 182.477546 48.376545 788.817093 0.143054 0.080363 0.063520 0.157186 3.082046 2.625092 3378.041452 1010.522498 841.572105 0.020085 11.882266
In [32]:
display(labels.shape) # Labels associated to each data point
display (labels.max())
display (labels.min())
         

(8950,)
7
0
In [33]:
y_kmeans = kmeans.fit_predict(creditcard_df_scaled)
display(y_kmeans)

array([0, 2, 6, ..., 7, 7, 7])
In [34]:
# concatenate the clusters labels to our original dataframe
creditcard_df_cluster = pd.concat([creditcard_df, pd.DataFrame({'cluster':labels})], axis = 1)
creditcard_df_cluster.head()
Out[34]:
BALANCE BALANCE_FREQUENCY PURCHASES ONEOFF_PURCHASES INSTALLMENTS_PURCHASES CASH_ADVANCE PURCHASES_FREQUENCY ONEOFF_PURCHASES_FREQUENCY PURCHASES_INSTALLMENTS_FREQUENCY CASH_ADVANCE_FREQUENCY CASH_ADVANCE_TRX PURCHASES_TRX CREDIT_LIMIT PAYMENTS MINIMUM_PAYMENTS PRC_FULL_PAYMENT TENURE cluster
0 40.900749 0.818182 95.40 0.00 95.4 0.000000 0.166667 0.000000 0.083333 0.000000 0 2 1000.0 201.802084 139.509787 0.000000 12 7
1 3202.467416 0.909091 0.00 0.00 0.0 6442.945483 0.000000 0.000000 0.000000 0.250000 4 0 7000.0 4103.032597 1072.340217 0.222222 12 1
2 2495.148862 1.000000 773.17 773.17 0.0 0.000000 1.000000 1.000000 0.000000 0.000000 0 12 7500.0 622.066742 627.284787 0.000000 12 2
3 1666.670542 0.636364 1499.00 1499.00 0.0 205.788017 0.083333 0.083333 0.000000 0.083333 1 1 7500.0 0.000000 864.206542 0.000000 12 7
4 817.714335 1.000000 16.00 16.00 0.0 0.000000 0.083333 0.083333 0.000000 0.000000 0 1 1200.0 678.334763 244.791237 0.000000 12 7
In [35]:
for i in creditcard_df.columns:
    plt.figure(figsize = (20,5))
    for j in range(8):
        plt.subplot(1,8,j+1)
        cluster = creditcard_df_cluster[creditcard_df_cluster['cluster'] == j]
        cluster[i].hist(bins = 20)
        plt.title('{}    \nCluster {} '.format(i,j))
  
    plt.show()

PRINCIPAL COMPONENT ANALYSIS¶

In [36]:
pca = PCA(n_components=2)
principal_comp = pca.fit_transform(creditcard_df_scaled)
display(principal_comp)

array([[-1.68222241, -1.0764406 ],
       [-1.13830449,  2.50651402],
       [ 0.96968867, -0.38355405],
       ...,
       [-0.92620335, -1.81078101],
       [-2.33655014, -0.65796128],
       [-0.55641626, -0.40049209]])
In [37]:
# Create a dataframe with the two components
pca_df = pd.DataFrame(data = principal_comp, columns =['pca1','pca2'])
pca_df.head()
Out[37]:
pca1 pca2
0 -1.682222 -1.076441
1 -1.138304 2.506514
2 0.969689 -0.383554
3 -0.873631 0.043177
4 -1.599435 -0.688573
In [38]:
# Concatenate the clusters labels to the dataframe
pca_df = pd.concat([pca_df,pd.DataFrame({'cluster':labels})], axis = 1)
display(pca_df)
pca1 pca2 cluster
0 -1.682222 -1.076441 7
1 -1.138304 2.506514 1
2 0.969689 -0.383554 2
3 -0.873631 0.043177 7
4 -1.599435 -0.688573 7
... ... ... ...
8945 -0.359629 -2.016141 6
8946 -0.564367 -1.639127 6
8947 -0.926203 -1.810781 6
8948 -2.336550 -0.657961 6
8949 -0.556416 -0.400492 6

8950 rows × 3 columns

In [39]:
pca_df.value_counts(pca_df.cluster)
Out[39]:
cluster
7    2718
0    2011
3    1181
2    1106
1     866
6     622
4     416
5      30
dtype: int64
In [40]:
plt.figure(figsize=(20,10))
ax = sns.scatterplot(x="pca1", y="pca2", hue = "cluster", data = pca_df, palette =['red','green','blue','pink','yellow','gray','purple', 'black'])
plt.show()

APPLY AUTO ENCODERS¶

In [41]:
from tensorflow.keras.layers import Input, Add, Dense, Activation, ZeroPadding2D, BatchNormalization, Flatten, Conv2D, AveragePooling2D, MaxPooling2D, Dropout
from tensorflow.keras.models import Model, load_model
from tensorflow.keras.initializers import glorot_uniform  # This is normalizer
from keras.optimizers import SGD

encoding_dim = 7

input_df = Input(shape=(17,)) # 17 Features 


# Glorot normal initializer (Xavier normal initializer) draws samples from a truncated normal distribution 

x = Dense(encoding_dim, activation='relu')(input_df)
x = Dense(500, activation='relu', kernel_initializer = 'glorot_uniform')(x)
x = Dense(500, activation='relu', kernel_initializer = 'glorot_uniform')(x)
x = Dense(2000, activation='relu', kernel_initializer = 'glorot_uniform')(x)

encoded = Dense(10, activation='relu', kernel_initializer = 'glorot_uniform')(x)

x = Dense(2000, activation='relu', kernel_initializer = 'glorot_uniform')(encoded)
x = Dense(500, activation='relu', kernel_initializer = 'glorot_uniform')(x)

decoded = Dense(17, kernel_initializer = 'glorot_uniform')(x)

# autoencoder
autoencoder = Model(input_df, decoded)

#encoder - used for our dimention reduction
encoder = Model(input_df, encoded)

autoencoder.compile(optimizer= 'adam', loss='mean_squared_error')
In [42]:
display (creditcard_df_scaled.shape)

(8950, 17)
In [43]:
autoencoder.fit(creditcard_df_scaled, creditcard_df_scaled, batch_size = 128, epochs = 25,  verbose = 1)

Epoch 1/25
70/70 [==============================] - 6s 56ms/step - loss: 0.5177
Epoch 2/25
70/70 [==============================] - 4s 55ms/step - loss: 0.2855
Epoch 3/25
70/70 [==============================] - 4s 56ms/step - loss: 0.2301
Epoch 4/25
70/70 [==============================] - 4s 56ms/step - loss: 0.2059
Epoch 5/25
70/70 [==============================] - 4s 55ms/step - loss: 0.1784
Epoch 6/25
70/70 [==============================] - 4s 55ms/step - loss: 0.1590
Epoch 7/25
70/70 [==============================] - 4s 57ms/step - loss: 0.1439
Epoch 8/25
70/70 [==============================] - 4s 55ms/step - loss: 0.1253
Epoch 9/25
70/70 [==============================] - 4s 55ms/step - loss: 0.1150
Epoch 10/25
70/70 [==============================] - 4s 55ms/step - loss: 0.1067
Epoch 11/25
70/70 [==============================] - 4s 55ms/step - loss: 0.1011
Epoch 12/25
70/70 [==============================] - 4s 55ms/step - loss: 0.0952
Epoch 13/25
70/70 [==============================] - 4s 57ms/step - loss: 0.0918
Epoch 14/25
70/70 [==============================] - 4s 60ms/step - loss: 0.0892
Epoch 15/25
70/70 [==============================] - 4s 58ms/step - loss: 0.0837
Epoch 16/25
70/70 [==============================] - 4s 56ms/step - loss: 0.0784
Epoch 17/25
70/70 [==============================] - 4s 56ms/step - loss: 0.0726
Epoch 18/25
70/70 [==============================] - 4s 55ms/step - loss: 0.0701
Epoch 19/25
70/70 [==============================] - 4s 63ms/step - loss: 0.0723
Epoch 20/25
70/70 [==============================] - 4s 60ms/step - loss: 0.0617
Epoch 21/25
70/70 [==============================] - 4s 58ms/step - loss: 0.0586
Epoch 22/25
70/70 [==============================] - 4s 57ms/step - loss: 0.0572
Epoch 23/25
70/70 [==============================] - 4s 57ms/step - loss: 0.0554
Epoch 24/25
70/70 [==============================] - 4s 55ms/step - loss: 0.0566
Epoch 25/25
70/70 [==============================] - 4s 55ms/step - loss: 0.0517
Out[43]:
<keras.callbacks.History at 0x201b02968e0>
In [44]:
autoencoder.summary()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 17)]              0         
                                                                 
 dense (Dense)               (None, 7)                 126       
                                                                 
 dense_1 (Dense)             (None, 500)               4000      
                                                                 
 dense_2 (Dense)             (None, 500)               250500    
                                                                 
 dense_3 (Dense)             (None, 2000)              1002000   
                                                                 
 dense_4 (Dense)             (None, 10)                20010     
                                                                 
 dense_5 (Dense)             (None, 2000)              22000     
                                                                 
 dense_6 (Dense)             (None, 500)               1000500   
                                                                 
 dense_7 (Dense)             (None, 17)                8517      
                                                                 
=================================================================
Total params: 2,307,653
Trainable params: 2,307,653
Non-trainable params: 0
_________________________________________________________________
In [45]:
autoencoder.save_weights('autoencoder.h5')
In [46]:
pred = encoder.predict(creditcard_df_scaled)
display (pd.DataFrame(pred))

280/280 [==============================] - 2s 8ms/step
0 1 2 3 4 5 6 7 8 9
0 0.000000 0.457111 0.890344 1.717090 0.967631 1.056988 0.0 0.0 0.000000 0.000000
1 1.246969 2.804320 2.092428 1.288674 1.347486 2.334675 0.0 0.0 0.057597 0.000000
2 0.740157 0.416786 0.131792 0.308030 1.390511 0.371790 0.0 0.0 1.168417 0.766312
3 0.174244 1.038864 1.484888 0.739221 2.466419 0.986579 0.0 0.0 0.394535 0.000000
4 0.000000 0.711937 0.688485 1.269932 0.978676 1.060225 0.0 0.0 0.000000 0.000000
... ... ... ... ... ... ... ... ... ... ...
8945 0.064023 0.380342 0.521539 0.457458 0.077806 1.276789 0.0 0.0 0.094928 1.314105
8946 0.023236 0.158743 0.112626 0.606525 0.384708 1.096888 0.0 0.0 0.255362 1.441767
8947 0.000000 0.241459 0.576213 0.383459 0.225735 1.201463 0.0 0.0 0.139119 1.057619
8948 0.136915 0.013083 0.560873 0.189760 0.429158 1.643312 0.0 0.0 0.014472 0.058017
8949 0.087425 0.000000 0.407008 0.000000 0.697960 1.725392 0.0 0.0 1.392876 1.378196

8950 rows × 10 columns

In [47]:
pred.shape
Out[47]:
(8950, 10)
In [48]:
scores_2 = []

range_values = range(1, 20)

for i in range_values:
    kmeans = KMeans(n_clusters= i)
    kmeans.fit(pred)
    scores_2.append(kmeans.inertia_)

plt.plot(scores_2, 'bx-')
plt.title('Finding right number of clusters')
plt.xlabel('Clusters')
plt.ylabel('scores') 
plt.show()
In [49]:
plt.plot(scores_1, 'bx-', color = 'r')
plt.plot(scores_2, 'bx-', color = 'g')
Out[49]:
[<matplotlib.lines.Line2D at 0x201ae3dfc40>]
In [50]:
kmeans = KMeans(4)
kmeans.fit(pred)
labels = kmeans.labels_
y_kmeans = kmeans.fit_predict(creditcard_df_scaled)
In [51]:
display(labels.shape) # Labels associated to each data point
display (labels.max())
display (labels.min())

(8950,)
3
0
In [52]:
df_cluster_dr = pd.concat([creditcard_df, pd.DataFrame({'cluster':labels})], axis = 1)
display(df_cluster_dr.head())
BALANCE BALANCE_FREQUENCY PURCHASES ONEOFF_PURCHASES INSTALLMENTS_PURCHASES CASH_ADVANCE PURCHASES_FREQUENCY ONEOFF_PURCHASES_FREQUENCY PURCHASES_INSTALLMENTS_FREQUENCY CASH_ADVANCE_FREQUENCY CASH_ADVANCE_TRX PURCHASES_TRX CREDIT_LIMIT PAYMENTS MINIMUM_PAYMENTS PRC_FULL_PAYMENT TENURE cluster
0 40.900749 0.818182 95.40 0.00 95.4 0.000000 0.166667 0.000000 0.083333 0.000000 0 2 1000.0 201.802084 139.509787 0.000000 12 0
1 3202.467416 0.909091 0.00 0.00 0.0 6442.945483 0.000000 0.000000 0.000000 0.250000 4 0 7000.0 4103.032597 1072.340217 0.222222 12 2
2 2495.148862 1.000000 773.17 773.17 0.0 0.000000 1.000000 1.000000 0.000000 0.000000 0 12 7500.0 622.066742 627.284787 0.000000 12 0
3 1666.670542 0.636364 1499.00 1499.00 0.0 205.788017 0.083333 0.083333 0.000000 0.083333 1 1 7500.0 0.000000 864.206542 0.000000 12 0
4 817.714335 1.000000 16.00 16.00 0.0 0.000000 0.083333 0.083333 0.000000 0.000000 0 1 1200.0 678.334763 244.791237 0.000000 12 0
In [53]:
pca = PCA(n_components=2)
prin_comp = pca.fit_transform(pred)
pca_df = pd.DataFrame(data = prin_comp, columns =['pca1','pca2'])
display (pca_df.head())
pca1 pca2
0 -1.055746 -0.363532
1 1.007257 -1.402239
2 -0.484735 0.673778
3 -0.125583 -0.421430
4 -1.085654 -0.393209
In [54]:
pca_df = pd.concat([pca_df,pd.DataFrame({'cluster':labels})], axis = 1)
pca_df.head()
Out[54]:
pca1 pca2 cluster
0 -1.055746 -0.363532 0
1 1.007257 -1.402239 2
2 -0.484735 0.673778 0
3 -0.125583 -0.421430 0
4 -1.085654 -0.393209 0
In [55]:
pca_df.value_counts(pca_df.cluster)
Out[55]:
cluster
0    6310
2    1573
1    1000
3      67
dtype: int64
In [56]:
plt.figure(figsize=(10,10))
ax = sns.scatterplot(x="pca1", y="pca2", hue = "cluster", data = pca_df, palette =['red','green','blue','yellow'])
plt.show()
In [ ]: